eduzhai > Applied Sciences > Engineering >

Better Than Reference In Low Light Image Enhancement Conditional Re-Enhancement Networks

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 10 pages

Abstract: Low light images suffer from severe noise, low brightness, low contrast, etc.In previous researches, many image enhancement methods have been proposed, butfew methods can deal with these problems simultaneously. In this paper, tosolve these problems simultaneously, we propose a low light image enhancementmethod that can combined with supervised learning and previous HSV (Hue,Saturation, Value) or Retinex model based image enhancement methods. First, weanalyse the relationship between the HSV color space and the Retinex theory,and show that the V channel (V channel in HSV color space, equals the maximumchannel in RGB color space) of the enhanced image can well represent thecontrast and brightness enhancement process. Then, a data-driven conditionalre-enhancement network (denoted as CRENet) is proposed. The network takes lowlight images as input and the enhanced V channel as condition, then it canre-enhance the contrast and brightness of the low light image and at the sametime reduce noise and color distortion. It should be noted that during thetraining process, any paired images with different exposure time can be usedfor training, and there is no need to carefully select the supervised imageswhich will save a lot. In addition, it takes less than 20 ms to process a colorimage with the resolution 400*600 on a 2080Ti GPU. Finally, some comparativeexperiments are implemented to prove the effectiveness of the method. Theresults show that the method proposed in this paper can significantly improvethe quality of the enhanced image, and by combining with other image contrastenhancement methods, the final enhancement result can even be better than thereference image in contrast and brightness. (Code will be available atthis https URL)

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×